Non-engineer - siamese intake manifold EFI 2

[hbtorana]

Good Day All,
Converting 1256cc Vauxhall Viva/Holden Torana to EFI. (sequential injection and sequential ignition)
Engine 4 cyl Siamese Intake port. Require help on manifold length.

Have GT-suite - Have access to SAE and articles and books.
In gt-suite have modelled the current intake/engine-exhaust - exisitng engine/manifold, and multiple other intake and exhausts variants. Crudely 3d modelled the conrod for centre of gravity etc info. Fed into the engine model.
Orignal engine with original manifold and exhaust dimension only approx 48% VE.

[Probaly going to end up modelling the carburettor in Gt-suite but given the lack of CUDA/FPGA only processor support apparently 100x cpu requirement for fluid simulation. Still put simulation time/run times inreach 30 mintues or so.]

Fudged about with the intake/exhaust lengths managed 82% VE, changed injection kg/hour rate and managed out to 92% VE@3200rpm - once off never kept the model.

The current manifold carby is semi-circular (center of port to center of port approx 22.78 inch length circumference), carby 45' angle from horizontal. 7.109 inch center intake port 1 to center of intake port 2.

Given the exisiting intake design (semi-circular) I would guess the rotunda effect (of sounds wave against a concave surface) would/could/should be used? Eg reflection/wave travels entire curved path length to other valves vs to the throttle body.
When all other valves closed (off the second intake port ) Second half of the semi-circle manifold would act as an approx 11 inch long helmholtz resonator.

I am an IT guy with computer science degree, now in Purchasing. I should ahve done Mechanical Engineering but IT was my calling at that particular point in life.

Ultimately I will 3d model then 3d print in skeleton in ABS (high temp epoxy/carbon fibre) or just have 3rd party 3d print the manifold or can 3d print then lost pla cast the manifold and throw it across a mill myself.

Have set up a home flow bench to get more accurate valve/head intake data. This is still in progress - require greater vacuum. Have calibration plates from another bench but not satisfied. Works but not happy.

Stuck with appropriate length initial guesstimates for intake and exhaust. Have modelled the exhaust based on race spec exhaust length and diameters from a UK site that has previously made race/rally exhausts for this engine.

All guidance appreciated.
I have already had a few people say give up, your not an engineer, I/We do not have time to teach you basic engineering.

Seen the other posts on runner length equations can push most of it in to excel and come up with numbers.
Just want someone with more knowledge than I to say look there at this.
This is where to start, this is the next step.

Thanks,

 

[BrianPetersen]

No personal experience with siamese intake ports BUT ...

I foresee fuel-distribution trouble because of those siamese intake ports if you try to inject fuel at the ports ("dry" intake manifold). You have a single pipe leading to two intake valves that draw sequentially from that single pipe (presumably with a single injector) and then a full crankshaft revolution of nothing. When do you plan to fire the injector? What happens to the fuel hanging out on the walls of the port (wall-wetting)?

If the injector is sized for the usual 80-ish-% duty cycle at full engine load then you have potentially nearly a full revolution of the injector loading up the port with fuel, a bunch of which gets drawn into the first cylinder in the sequence of that pair, and then the second one that follows right afterward gets starved of fuel.

If you fire it twice - once for each cylinder while that cylinder is drawing in - and keep the duty cycle down then you can kinda-sorta co-ordinate the fuel flow with the air flow, and make some sort of probably-huge correction for the fuel that lands on the port walls and stays there from that cycle into the next.

I've only ever tuned engines with the usual multi-point and individual runner intake designs, and even there, the port wall-wetting can be nuisance enough, without having to worry about each cylinder acting very substantially different from each other!

The details of splitting hairs about what the pulsations in the intake manifold are going to do aren't worth pursuing relative to how you plan to handle the fuel distribution.

The Austin A-series has the same issue. I know only as much about those as Google will find, like this:

https://www.minimania.com/A_Basic_Guide_to_Electronic_Fuel_Injection_for_Minis

It appears that if you want to use the 5-port head (siamese 1-2 and 3-4 inlets, siamese exhaust on cylinders 2-3) they advocate what amounts to throttle-body fuel injection with the injectors at the throttle body, not at the cylinders - basically replacing the carburetor.

https://www.minimania.com/part/SC0126/Classic-Mini-Fuel-Injection-Upgrade-Conversion-Kit

 

[dgallup]

There are plenty of twin spray injectors that with proper targeting may be able to get a squirt onto each intake valve without too much wall wetting. These are normally used to inject onto the twin intake valves of a single cylinder 4 valve head. You will need something with as wide a twin spray separation angle as possible. There were lots of early fuel injected engines that used what was called "simultaneous double fire". All the injectors fired at the same time, once per revolution. Injection timing was not optimal for any cylinder. Not as good as a sequential injection system but it works. The most important thing is injector targeting to get as much fuel as possible to hit the back of the hot intake valve.

----------------------------------------

The Help for this program was created in Windows Help format, which depends on a feature that isn't included in this version of Windows.

 

[gruntguru]

dgallup. I think your suggestion will still see more fuel flowing to the "first valve to open" of the siamese pair. Much of the fuel injected with valves closed will be drawn in by the first valve to open.

je suis charlie

 

[BrianPetersen]

Batch fire or simultaneous double fire worked ok (ish) with port fuel injection and individual runners per cylinder because (if the runners are decently long) there's essentially no chance of fuel crossing over from one cylinder to the other, and normal spark ignition engines aren't all that sensitive to injection timing because everything that goes into the cylinder ends up getting all mixed up inside the cylinder anyhow. It worked ok (ish) with single-point throttle body injection because those used a carburetor-style intake manifold that usually had a plenum underneath the throttle body that at least somewhat mixed up the air and fuel coming in through the throttle body before it went into each intake runner. Single-point injection still has all of the wet-manifold wall-wetting and fuel distribution issues that carburetors always had. Emission standards of the 1990s pretty much wiped out single-point injection and standards of the 2000s wiped out anything other than sequential multi-point injection (with an individual intake runner per cylinder - no more siamesed intake ports).

I just thought of another engine that started out with siamesed intake ports ... the original VW air-cooled flat-four. They went to dual-port heads sometime in the 1960s, and when they fitted multi-point fuel injection to that engine in the late 1960s, they redid the intake manifolds to use four completely separate (and equal-length) intake runners drawing air from a central plenum.

So ... we're left with finding out what the original poster's objectives are. My blunt recommendation is that if the objective is to get better performance, get a different engine. Maybe there's a place for period-correct hot-rodding of that engine, but fuel injection won't be part of that.

 

[hbtorana]

Hi,

From fuel point of view I am confidently set, just need to work out how to package the 4 injectors/rail/pressure reg/dampener into the limited space. Have sets of injectors low impedance 550cc,850cc, high impedance bosch 1000cc/1200cc.

550cc will get me somewhere to 60BHP. The 1000cc/1200cc injectors should get 130-140bhp I only need 80-90bhp and then amr500 charger later on.

Fully sequential 4 injectors, megasquirt extra siamese code.

http://www.jbperf.com/sequential/index.html

Cam/crank sensor, innovate lc-1 wideband o2 sensors,map sensors, fuel rail pressure sensor and few other pressure sensors for the intake (analog), along with EGT temp sensors for logging not feeding the megasquirt.

The ms extra code has the goodness for fully sequential siamese, covering the reduced injection window for cylinders 2/3 eg need to dump the entire cylinder fuel requirement in the reduced timing cam degree window where #1 valve closed & #2 open. (etc,etc).

I even want to tinker with electronic controlled EGR but megasquirt has no code for it. leave that for now and functionally model, develop schematic & prototype on an arm canbus dev board.

Head ports are currently 28.8mmm, can expand up to 34mm without getting in to water jacket or exhaust ports.
With the GT model after 31.1 mm approx. there was no VE improvement - probably due to rest of the intake manifold limiting flow.

I want to know about details of splitting hairs - about what the pulsations in the intake manifold are going to do. Lengths/diameters/plenum size.

Thanks.


I have another quesiton for eng-tips on ditching the throttle body completely and going to neural network/PID controlled variable length/diameter intake but will leave that for another conversation.

 

[BrianPetersen]

Hmm, interesting article in there on siamese operation. They're using one injector per siamesed port and recommend firing it once per revolution, 30 degrees ATDC on the second cylinder in the firing sequence (I suspect it will actually be RPM dependent and a little earlier at high revs due to time delay between firing the injector and the actual fuel spray actually making it to the intake valve), and sizing it for around 58% duty cycle at full load. The idea is to inject all of the fuel for the second cylinder in the firing sequence while that cylinder's intake valve is open, and then the fuel for the first cylinder in the next firing sequence can be injected with the valves closed, and it will just draw it in when it opens. I'd suggest having a separate lambda sensor for the inner two cylinders and another one for the outer two.

Your intake runners are only drawing one cylinder's worth of intake air at a time, so I'm thinking you would size them as if they were individual runners. For 20ish horsepower per cylinder, 30mm give or take at the face of the cylinder head looks about right. Don't go too big. Runners will probably want to be tapered from around 35-ish mm diameter at the plenum end to 30-ish to match the cylinder head face. No idea what your engine revs to, I'm guessing 6000-ish. It's probably going to want longer runners than what you've got space for. My ruff guess is it will want 400-ish mm measured from intake valve to plenum. This will be a "cut and try" ... whether in software or in metal ... probably in software followed by metal.

In a 1D simulation, don't worry about whether the runners are straight or gently curved ... just don't build it with sharp 90 degree corners. Measure length along the centerline of the ducts.

 

[RRiver]

@hbtorana

Interesting. Other than mentioning VE you really don't say what you're pursuing with your design. Performance, VE, economy, emissions, all the above?

I always wondered what Honda could have done with the CVCC design if their only focus hadn't been on emissions, economy and regulations? It may have been? Just never heard about it.

 

[hbtorana]

Apologies All for tardy time to respond.
Performance, VE, economy, emissions, all the above? All of the above.

Economy wise - Not sure if I could best the tiny little carby and 3.55 ratio diff - fully laden car - trundling along with the rest of the car club, around 6.5l/260km or so. (1/4 tank Melbourne to Echuca).

Newest engine in car - a 4agze for about 2 years.